Hi Emil, On Wed, 29 Aug 2018 at 17:33, Emil Velikov <emil.l.veli...@gmail.com> wrote: > On 29 August 2018 at 07:17, Daniel Stone <dani...@collabora.com> wrote: > > -static const struct sockaddr_un example_sockaddr_un; > > +static struct sockaddr_un example_sockaddr_un; > > > > #define TOO_LONG (1 + sizeof example_sockaddr_un.sun_path) > > > > @@ -69,6 +69,11 @@ TEST(socket_path_overflow_client_connect) > > d = wl_display_connect(path); > > assert(d == NULL); > > assert(errno == ENAMETOOLONG); > > + > > + /* This is useless, but prevents a warning about example_sockaddr_un > > + * being discarded from the compilation unit. */ > > + strcpy(example_sockaddr_un.sun_path, "happy now clang?"); > > + assert(example_sockaddr_un.sun_path[0] != '\0'); > > Why don't you add _attrubute__((used)) instead of doing the blame > game? We already use it in the repo.
This was already merged, but sure, a follow-up patch to do this would be fine. > Side note: the manpage says "sun_path[108]" and also points out that > "some implementations have sun_path as short as 92 bytes" > The check in wl_socket_init_for_display_name uses sizeof, yet prints a > "exceeds 108 bytes" message. > > I guess the message should be fixed? Yes, it looks like a follow-up patch to use sizeof would be correct as well. Cheers, Daniel _______________________________________________ wayland-devel mailing list wayland-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/wayland-devel
